Equivalent circuit parameter estimation of an induction machine using finite element analysis - DIMITRIOS SAGRIS

Page created by Julio Thompson
 
CONTINUE READING
Equivalent circuit parameter estimation of an induction machine using finite element analysis - DIMITRIOS SAGRIS
Equivalent circuit parameter estimation
of an induction machine using
finite element analysis

DIMITRIOS SAGRIS

DEPARTMENT OF ELECTRICAL ENGINEERING
C HALMERS U NIVERSITY     OF   T ECHNOLOGY
Gothenburg, Sweden 2021
www.chalmers.se
Equivalent circuit parameter estimation of an induction machine using finite element analysis - DIMITRIOS SAGRIS
Master’s thesis 2021

Equivalent circuit parameter estimation
    of an induction machine using
        finite element analysis

             DIMITRIOS SAGRIS

        Department of Electrical Engineering
       Division of Electric Power Engineering
      Chalmers University of Technology
              Gothenburg, Sweden 2021
Equivalent circuit parameter estimation of an induction machine using finite element
analysis
DIMITRIOS SAGRIS

© DIMITRIOS SAGRIS, 2021.

Supervisor: Torbjörn Thiringer, Department of Electrical Engineering
Examiner: Torbjörn Thiringer, Department of Electrical Engineering

Master’s Thesis 2021
Department of Electrical Engineering
Division of Electric Power Engineering
Chalmers University of Technology
SE-412 96 Gothenburg
Telephone +46 31 772 1000

Typeset in LATEX, template by Magnus Gustaver
Printed by Chalmers Reproservice
Gothenburg, Sweden 2021

Equivalent circuit parameter estimation of an induction machine using finite element
analysis
DIMITRIOS SAGRIS
Department of Electrical Engineering
Chalmers University of Technology

                                                                                  iii
Abstract
In this project, finite element analysis has been conducted in order to estimate the
parameters of an induction machine. A model of a 15kW induction machine in An-
sys Maxwell was the main object of the investigation. Using the field analysis from
Ansys and in combination with the Ansys result outputs, the stator leakage induc-
tance and the rotor resistance were investigated. After that, a parameter estimation
of the parameters in the τ model and in the k was carried out.

The performed analysis showed that there can be a good estimation of the stator
leakage inductance in the case of having a linear material in the core. Moreover,
the estimation of the rotor resistance using the field analysis from Ansys gave a
very close value compared to the theoretically calculated value of the rotor resis-
tance (from geometrical data). Finally, considering the rotor resistance, the stator
flux from Ansys result output and the stator current from Ansys result output, a
parameter estimation process could be performed.

Keywords: induction machine, parameter calculation, finite element analysis.

iv
Acknowledgements
I would like to thank my supervisor and examiner, professor Torbjörn Thiringer, for
the encouraging guidance and vital advice.

I would like to thank PhD student Meng-Ju Hsieh, for the nice cooperation.

                                          Dimitrios Sagris, Gothenburg, June 2021

                                                                                 v
Contents

1 Introduction                                                                                                                         1
  1.1 Background . . .      . . . . . .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .    1
  1.2 Related work . .      . . . . . .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .    1
  1.3 Scope . . . . . .     . . . . . .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .    1
  1.4 Aim and layout of     the thesis    .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .   .    1

2 Theoretical background                                                                                                               3
  2.1 Induction machine theory . . . . . . . . . . . . . . . . . . . . . . . .                                                         3
  2.2 Inductance and flux connection . . . . . . . . . . . . . . . . . . . . .                                                         8
  2.3 Parameter calculation . . . . . . . . . . . . . . . . . . . . . . . . . . .                                                      9

3 Case setup                                                                                                                          12

4 Analysis                                                                                                                            16
  4.1 Inductance investigation . . . . . . . . . . . .                            .   .   .   .   .   .   .   .   .   .   .   .   .   16
      4.1.1 Flux distribution . . . . . . . . . . . .                             .   .   .   .   .   .   .   .   .   .   .   .   .   16
      4.1.2 Finding Leakage Inductance . . . . . .                                .   .   .   .   .   .   .   .   .   .   .   .   .   25
      4.1.3 Adding saturation . . . . . . . . . . . .                             .   .   .   .   .   .   .   .   .   .   .   .   .   31
  4.2 Power balance check . . . . . . . . . . . . . .                             .   .   .   .   .   .   .   .   .   .   .   .   .   37
  4.3 Rotor resistance . . . . . . . . . . . . . . . . .                          .   .   .   .   .   .   .   .   .   .   .   .   .   43
      4.3.1 Rotor resistance from locked rotor test                               .   .   .   .   .   .   .   .   .   .   .   .   .   44
      4.3.2 Rotor resistance from field analysis . .                              .   .   .   .   .   .   .   .   .   .   .   .   .   48
  4.4 Stator leakage inductance . . . . . . . . . . .                             .   .   .   .   .   .   .   .   .   .   .   .   .   51
  4.5 Parameter estimation . . . . . . . . . . . . . .                            .   .   .   .   .   .   .   .   .   .   .   .   .   52
      4.5.1 Processing the data . . . . . . . . . . .                             .   .   .   .   .   .   .   .   .   .   .   .   .   52
      4.5.2 Results at load test . . . . . . . . . . .                            .   .   .   .   .   .   .   .   .   .   .   .   .   54
  4.6 Sustainable aspects . . . . . . . . . . . . . . .                           .   .   .   .   .   .   .   .   .   .   .   .   .   57

5 Conclusion and future work                                                       58
  5.1 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
  5.2 Future work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59

                                                                                                                                      vii
1
                           Introduction

1.1      Background
Identifying the parameters of an induction machine is crucial for the design of an
effective controller. The temperature variations affect the conductivity of a material.
The magnetic field can affect the permeability of a material. This is one reason why
the leakage and magnetization inductance of an induction machine are not constant
during different operation points.

1.2      Related work
Some models of parameter identification considering magnetic saturation have been
found in literature where different methods in order to achieve inductance estima-
tion have been described. In [1] the use of cost function is combined with inductance
estimation. In [12] an iteration process is described in order to estimate the leakage
and magnetizing inductance. In [2] the magnetizing inductance is expressed consid-
ering a constant leakage inductance.

1.3      Scope
All investigations were made using finite element analysis. A 15kW induction ma-
chine using a squired cage rotor was simulated in Ansys Maxwell.

1.4      Aim and layout of the thesis
The purpose of this project is to investigate the parameter estimation of an induc-
tion machine using finite element analysis.

                                                                                     1
1. Introduction                                                                  2

In chapter 4.1 the outputs of the inductance matrix from Ansys Maxwell are inves-
tigated. In chapter 4.2 the power balance is examined and a manual calculation of
the torque is described. In chapter 4.3 the rotor resistance is estimated with two
different approaches. In chapter 4.4 the stator leakage inductance is approximated.
In chapter 4.5 a parameter investigation is described from load test simulations.

2
2
             Theoretical background

2.1     Induction machine theory
In this thesis work an induction machine using a squired cage rotor is examined in
steady state. Equations of the τ model and equations of the k model are used in
steady state. More models of the induction machine can be found in [2].

Induction machine in xy coordinates, τ model

In a rotating system (xy coordinates) and with a speed of ω1 the equations of the
stator voltage, rotor voltage, stator flux, rotor flux and electromagnetic torque are

                                            dψs
                             us = Rs is +       + jω1 ψs                        (2.1)
                                             dt
                                         dψr
                          ur = Rr ir +       + j(ω1 − ωr )ψr                    (2.2)
                                          dt
                                 ψs = Ls is + Lm ir                             (2.3)
                                 ψr = Lr ir + Lm is                             (2.4)
                                    3np
                             Te =       (ψsx isy − ψsy isx )                    (2.5)
                                     2
where
us stator voltage in xy coordinates
is stator current in xy coordinates
Rs stator resistance
Rr rotor resistance
ψs stator flux
ψr rotor flux
ω1 synchronous electrical rotation speed
ωr rotor electrical rotation speed
Ls self inductance of the stator
Lr self inductance of the rotor
Lm mutual inductance between stator and rotor
Te electromagnetic torque

                                                                                   3
Theoretical background                                                               4

IM in xy coordinates, k model

Introducing b as an important ratio constant,

                                             Lm
                                      b=                                        (2.6)
                                             Lr
and according to [7] the units of the rotor flux and rotor current in the k model can
be related as

                                     ΨR = bψr                                   (2.7)
                                               ir
                                        iR =                                    (2.8)
                                               b

The magnetization inductance and leakage inductance in the k model can expressed
as

                                            L2m
                                     LM   =                                     (2.9)
                                            Lr
                                   Lσ = Ls − LM                                (2.10)
and the rotor resistance in the k model is

                                     RR = b2 Rr                                (2.11)
while the stator flux and the rotor flux in the k model can expressed as

                             ψs = (Lσ + LM )is + LM iR                         (2.12)

                                ψR = LM is + LM iR                             (2.13)
where
Lr self inductance of the rotor
ΨR rotor flux in the k model
iR rotor current in the k model
LM magnetization inductance
Lσ leakage inductance in the k model

From (2.12) and (2.13), the stator flux and rotor current can be expressed also as

                                  ψs = L σ i s + Ψ R                           (2.14)
                                        ΨR − LM is
                                 iR =                                          (2.15)
                                           LM

4
Theoretical background                                                                5

Now the equations of the stator voltage, rotor voltage and torque become

                                             dψs
                              us = Rs is +       + jω1 ψs                       (2.16)
                                              dt
                                        dΨR
                       ur = 0 = RR iR +        + j(ω1 − ωr )ΨR                  (2.17)
                                         dt
                                        3np
                                  Te =      =ψs∗ is                             (2.18)
                                         2
                                  3np
                             Te =     (ΨRd isq − ΨRq isd )                      (2.19)
                                   2

In dq coordinates the rotor flux is

                                 ΨR =
Theoretical background                                                            6

then taking the real part of (2.22)

                                                     RR
                                    RR isd = ΨR                               (2.26)
                                                     LM
                                       ΨR = isd LM                            (2.27)
and taking the imaginary part of (2.22), we find

                                                        RR isq
                               ωslip = (ω1 − ωr ) =                           (2.28)
                                                         ΨR
where
ωslip angular velocity of the slip frequency

Considering (2.28) and (2.23), torque can be expressed as

                                           3np RR i2q
                                      Te =                                    (2.29)
                                            2 wslip

Losses

The induction machine in this paper is examined without taking into account me-
chanical friction. The windings of the machine are made of copper and the rotor bars
are made of aluminium. The total losses considering only electromagnetic analysis
will be

                            Ptotallosses = Pcu + Psolid + P f e               (2.30)
where copper looses are
                                           √     2
                                   Pcu =       3IsRM S Rs                     (2.31)
and solid losses are
                                             √     2
                                  Psolid =       3IrRM S Rr                   (2.32)
where
Pcu      total copper losses of the stator windings
Pf e     total iron losses
Psolid    total losses of the rotor bars
IsRM S    RMS value of stator current
IrRM S     RMS value of the equivalent in stator side rotor current

6
Theoretical background                                                               7

Iron losses can be expressed with different models as explained in [4]. One model is
the separation of iron losses in hysteresis based losses (Phys ) and eddy current based
losses (Peddy ).

                                   Pf e = Phys + Peddy                          (2.33)

According to [4] (2.33) lacks accuracy for Si-Fe alloys and a third component in the
iron losses can be added called excess losses. Iron losses can be re-expressed as

                               Pf e = Phys + Peddy + Pexcess                    (2.34)
where
Pexcess is the excess losses

Losses can be divided to more categories which can be found in [8].

Power balance

Power balance should be satisfied, which means that the input electrical power
should be equal to the output power. The output power is comprised by the me-
chanical power and the losses of the machine

                                        Pin = Pout                              (2.35)
                         ua ia + ub ib + uc ic = T Ωr + Ptotallosses            (2.36)
where
Pin total input electrical power
Pout total output power
T output Torque
ua voltage of phase a
ub voltage of phase b
uc voltage of phase c
ia current of phase a
ib current of phase b
ic current of phase c
Ωr mechanical speed of the rotor

                                                                                     7
Theoretical background                                                              8

2.2     Inductance and flux connection
The flux linkage of phase a caused by the influence of the 3 stator currents isa , isb
and isc is given by

                       Ψsa = (Lsσ + Lh )isa + Mab isb + Mac isc                 (2.37)
where
Ψsa the flux of phase a
Lsσ stator leakage inductance
Lh main inductance of a single phase
Mab mutual inductance between phase a and b
Mac mutual inductance between phase a and c

In case of a star connection the summation of the three stator currents is zero

                                  isa + isb + isc = 0                           (2.38)

If the mutual inductances are equal, then

                                     Mab = Mac                                  (2.39)

Combing (2.38) and (2.39), equation (2.37) becomes

                            Ψsa = (Lsσ + Lh )isa + M isa                        (2.40)

The stator flux of the phase a can be also expressed as

                                     Ψsa = Ls isa                               (2.41)
where Ls is the self inductance

M and Lh are connected with an angle θ as

                                    M = Lh cosθ                                 (2.42)
where θ is the angle between 2 phase windings.

8
Theoretical background                                                            9

2.3     Parameter calculation
Stator resistance

The resistivity of copper can be expressed as

                   ρCu = 16.8 · 10−9 · (1 + 3.9 · 10−3 · (Tcu − 20))          (2.43)
where
Tcu temperature of the copper in Celsius

The resistance of the active winding is
                                       ρCu · LAW · ns qs
                              Rsaw =                                          (2.44)
                                       Acu · cs · nstrands
where
ρCu is the resistivity of copper
LAW is the length of the active winding
ns is the number of turns per coil
qs is the number of slots per phase per pole pair
Acu is the area of the copper conductor
cs is the winding factor which depends on the way the phases are connected and the
parallel branches
nstrands is the number of strands

The resistance of the end winding is
                                       ρCu · LEW · ns · qs
                             Rsew =                                           (2.45)
                                        Acu cs · nstrands
where
LEW is the length of the end winding

The total stator resistance is comprised of the resistance of the active winding and
the resistance of the end winding

                                 Rs = Rsaw + Rsew                             (2.46)

It should be noted that in this project the simulations are been carried out in 2D
simulation using FEA and the end winding resistance is not contributing to the
stator resistance.

                                                                                  9
Theoretical background                                                               10

Rotor resistance

The resistivity of the rotor bar in case of aluminium material can be expressed as

                    ρal = 28.2 · 10−9 · (1 + 4.3 · 10−3 · (Tal − 20))            (2.47)
where
Tal is the temperature of the aluminium in Celcius

The resistance of a single bar is

                                    Rbar = ρal La /Abar                          (2.48)
where
La is the length of the bar
Abar is the cross section of the bar in the case of current in z direction

The resistance of the end ring is

                        Rendring = (2π · ρal · rscr )/(Qr · Ascr )               (2.49)
where
Ascr can be expressed according to [5] as
                                              Qr Abar
                                     Ascr =                                      (2.50)
                                              πpoles
where
Qr is the number of total bars
poles is the number of total poles
rscr is the mean radius of the end ring

Expressing the rotor units to stator equivalent units, the coefficients for this conver-
sion are obtained. The coefficient to convert the rotor current to stator equivalent
rotor current is

                                        3n2s qs2 k12 poles2
                              coefreqS =                                      (2.51)
                                               Qr c2s
and the coefficient to convert the rotor resistance to stator equivalent rotor resis-
tance is
                                                 Qr cs
                               coefieqS =                                        (2.52)
                                            3ns qs k1 poles
where
k1 is the harmonic winding factor

10
Theoretical background                                                               11

coefreqS is the coefficient to connect the rotor equivalent resistance to stator equiv-
alent
coefieqS is the coefficient to connect the rotor equivalent current to stator equivalent

The stator equivalent resistance of the end ring is
                                      coefreqS · 2 · Rendring · Qr
                        Rrendring =                                              (2.53)
                                             (π · poles)2

The stator equivalent rotor bar resistance is

                                 Rrbar = coefreqS Rbar                           (2.54)

The total rotor resistance of the rotor in the stator side can be expressed as

                                Rr = Rrbar + Rrendring                           (2.55)

                                                                                     11
3
                             Case setup

In this project all investigations were made using finite element analysis. A 15kW
induction machine using a squired cage rotor was simulated in Ansys Maxwell which
was the main object of investigation. However, some more modified versions were
made in order to investigate inductances as described in chapter 4.

Induction Machine 15kW

Table 3.1 shows the parameters of the stator geometry. Table 3.2 shows the pa-
rameters of the rotor geometry. Figure 3.1 shows the position of the geometrical
parameters in the stator and rotor slot.

The number of turns per coil are 38 and a single layer of winding is modeled per
stator slot. One winding per pole pair is comprised by 2 coils connected in series.
The total number of parallel branches per phase are 3. All three windings had a
star connection with each other. No end ring is modeled in the model.

Stator windings are made of copper and rotor bars are made of aluminum. M700
is the non-linear material of the core. Table 3.3 shows the basic characteristics of
these three materials. Table 3.4 shows the points of the B-H curve of M700.

It should be noted that iron losses exist in the core material. However, the simula-
tion settings are modified in a way that iron losses do not affect the field solutions
of Ansys Maxwell.

12
Case setup                                                                   13

Table 3.1: Stator geometry

      Name       Value       Description
      SOD        291.2mm     Stator outer dimameter
      SID        190.2mm     Stator inner diameter
      slots      36          Number of stator slots
      HSO        1mm         Slot opening heght
      HSO1       0           Slot closed bridge heght
      HS1        1mm         Slot wedge heght
      HS2        13.5mm      Slot body heght
      BSO        3.5mm       Slot opening width
      BS1        8.5mm       Slot wedge maximum width
      BS2        11mm        Slot body bottom width, 0 for parallel teeth
      Rs         2mm         Slot body bottom fillet

Table 3.2: Rotor goemetry

      Name      Value        Description
      ROD       189.3mm      Rotor outer dimameter
      RID       55mm         Rotor inner diameter
      slots     39           Number of rotor bars
      HSO       8mm          Slot opening heght
      HSO1      0            Slot closed bridge heght
      HS1       0.2mm        Slot wedge heght
      HS2       9.7mm        Slot body heght
      BSO       3.35mm       Slot opening width
      BS1       5.9mm        Slot wedge maximum width
      BS2       3.9mm        Slot body bottom width, 0 for parallel teeth
      Rs        0.3mm        Slot body bottom fillet

Table 3.3: Materials used for IM

 Material                    Conductivity (S/m)            Relative permebility
 Copper                      57350900                      1
 Aluminium                   33055600                      1
 M700                        0                             B-H curve

                                                                             13
Case setup                                                           14

Figure 3.1: Example of a stotor slot(left) and a rotor slot(right)

Table 3.4: Points for the B-H curve of M700
                            H (A/m)       B (T)
                            0             0
                            67.8          0.1
                            88.3          0.2
                            99.2          0.3
                            108           0.4
                            116           0.5
                            124           0.6
                            132           0.7
                            142           0.8
                            152           0.9
                            164           1
                            180           1.1
                            206           1.2
                            254           1.3
                            363           1.4
                            690           1.5
                            1760          1.6
                            4230          1.7
                            8130          1.8
                            8239          1.802
                            8571          1.81
                            9137          1.82
                            9955          1.83
                            11060         1.85
                            12504         1.86
                            14381         1.875
                            16878         1.887
                            20467         1.899
                            30000         1.9193
                            150000        3.8

14
Case setup                                                                                      15

Figure 3.2 shows the model of the induction machine in Ansy Maxwell. Some times
the induction machine was excited with voltage excitation and some times with cur-
rent excitation.

When current excitation was applied a typical simulation was set to last 1 second.
When voltage excitation was applied a typical simulation was set to last 0.5 seconds.

A simulation with voltage excitation typically reaches steady state after 0.4 seconds.
One extra resistance was added (in voltage excitation case) in order to damp the
high initial currents. Figure 3.3 shows how this extra resistance looks in time.

Figure 3.2: IM main model

                                                         Rextra in time
                                    50

                                    45                                             Rextra
                                    40

                                    35
                     Rextra [ohm]

                                    30

                                    25

                                    20

                                    15

                                    10

                                    5

                                    0
                                         0   0.5   1   1.5      2        2.5   3     3.5    4
                                                             time [ms]

Figure 3.3: Extra resistance added as an additional stator resistance

                                                                                                15
4
                                 Analysis

4.1      Inductance investigation

Some efforts have been made in order to calculate stator inductances from the An-
sys simulations. It should be mentioned that Ansys Maxwell provides an inductance
matrix calculation option which is used in the simulations. However, it is unclear
what the results from the Ansys inductance matrix calculation stand for and how
useful these results are for our investigation.

4.1.1     Flux distribution

In this part flux distribution will be examined. First, it is desired to eliminate leak-
age flux and try to make the flux distribution homogeneous. In order to achieve the
first one, the slot opening is made from an unreal material with µr of 0.0001. In
order to achieve the second one, the slot opening is modified as shown in the Figure
4.1 and the core material is replaced by a steel with constant relative permeability
of 10000. Moreover, the rotor bars are made of the new core material for this initial
study.

A modified geometry as shown in the Figure 4.3 is tested with a DC current of 50A
and no rotor speed. Table 4.1 shows the total flux given by Ansys result output, the
main flux that comes from the field calculator and the flux through the slot opening
that comes from the field calculator. Moreover, Table 4.1 shows two cases of slot
material.

16
Inductance investigation                                                     17

Figure 4.1: Modified slot opening, light green is the unreal material with µr of
0.0001

The flux from the field analysis can be calculated from

                                  ZZ
                           Ψf =            Bnormal dS · Nturnscoil         (4.1)
                                       S

where
Ψf is the fundamental flux
Bnormal is the normal component of the flux density
S is the coss section of the Bnormal
Nturnscoil is the number of turns per coil

The line of the stator surface is been taken according to Figure 4.2

Figure 4.2: Line of stator surface

                                                                             17
Inductance investigation                                                          18

Figure 4.3: Modified geometry

Table 4.1: Total and main flux of the coil taken from Ansys result output and
from Ansys field calculator

     Material of slot   Main flux com-       Flux    coming   Flux      through
     opening            ing from the An-     from the Ansys   one slot opening
                        sys field calcula-   result  output   (Wb)       coming
                        tor (Wb)             (Wb)             from the Ansys
                                                              field calculator
     µr = 0.0001        1.108                1.108            5.6 · 10−5
     air                1.102                1.882            0.3935

18
Inductance investigation                                                              19

Two more cases are examined with 2 different coils as shown in Figure 4.4 and Figure
4.5. In the first case, the coil occupies half of the total geometrical angle of the pole
pair while in the second case the coil occupies 2/3 of the total geometrical angle of
the pole pair. The total flux density inside the coil will be the same as the total flux
density outside the coil

                           Z                    Z
                               Bnormalin dS =       Bnormalout dS                  (4.2)

and in the case of homogeneous flux distribution the flux density is the same every-
where. Equation (4.2) will result to

                        Bnormalin Acrossin = Bnormalout Acrossout                  (4.3)

where
Bnormalin is the normal flux density between the coil conductors
Bnormalout is the normal flux density outside the coil
Acrossin is the cross section of the Bnormalin
Acrossout is the cross section of the Bnormalout

In Figure 4.6 and Figure 4.7 are shown the Bnormal along the stator surface for the
examined cases respectively. It is clear that when Acrossout is smaller, then Bnormalout
becomes bigger according to (4.3). The spots where Bnormal is reaching 0 quickly
indicate the positions of the slot openings. One more visualization of the magnetic
flux density vectors is shown in the Figure 4.8 for the second case only.

Figure 4.4: Position of the coil in the first examined case

                                                                                      19
Inductance investigation                                                            20

Figure 4.5: Position of the coil in the second examined case

Figure 4.6: normal flux density across the stator surface for the first examined
case

Figure 4.7: Normal flux density across the stator surface for the second examined
case, here Acrossout is 3 times smaller than Acrossin resulting in that Bnormalout is 3
times bigger

20
Inductance investigation                                                         21

Figure 4.8: Normal flux density vectors in the area of the rotor for the second
examined case

Now it is desired to find the inductance in the case of 3 coils excited by 3 phases.
The modified version is shown in Figure 4.9.

Figure 4.9: Modification with 3 coils and 3 phase excitation

First, the influence of one coil only as shown in Figure 4.10 will be studied. A
simulation is performed with a current of 50A RMS, frequency of 50Hz and no rotor
speed. The results are shown in Table 4.2

Figure 4.10: Examining the desired modification with the influence of only 1 coil.

                                                                                 21
Inductance investigation                                                           22

In the case that the coil of phase a is examined

                                    Acrossout = Acrossin                      (4.4)

According to (4.3) it is obtained

                               Bnormalin = Bnormalout                         (4.5)
and the flux that flows between two slots, cause by phase a, can be expressed as
                             1                    1
                        ΨAA = Bnormalin Acrossin = ΨcoilA                     (4.6)
                             3                    3
where
ΨAA is the flux flowing between 2 slots caused only by the coil of phase a
ΨcoilA is the total flux of the coil of phase a

In the case of 3 coils excited by 3 phase currents as shown in Figure 4.9, it can
similarly obtained that the flux that flows between two slots, cause by phase b and
c can be expressed as
                                          1
                                     ΨBB = ΨcoilB                             (4.7)
                                          3
                                          1
                                     ΨCC = ΨcoilC                             (4.8)
                                          3
where
ΨBB is the flux flowing between 2 slots caused only by the coil of phase b
ΨcoilB is the total flux of the coil of phase b
ΨCC is the flux flowing between 2 slots caused only by the coil of phase c
ΨcoilC is the total flux of the coil of phase c

Table 4.2 shows the influence of phase a, phase b and phase c in all the teeth that
are between the beginning and the end of the coil of phase a

22
Inductance investigation                                                          23

Table 4.2: All fluxes that will build via superposition the total flux of phase a in
the case where all 3 coils are excited. MBA means the mutual flux between phase b
and phase a, while MCA means the mutual flux between phase c and phase a.

                          Fluxes from slot A+ to slot A-
                                         tooth (slotx slotxx)
    Flux   type         A+ C-              C- B+                B+ A-
    Flux   Phase A      ΨAA                ΨAA                  ΨAA
    Flux   MBA          -ΨBB               -ΨBB                 ΨBB
    Flux   MCA          ΨCC                -ΨCC                 −ΨCC

Next step is to apply superposition and find the total flux of phase a. The flux of
phase a can be expressed as

                            ΨA = 3ΨAA − 1ΨBB − 1ΨCC                             (4.9)

Table 4.3 shows the results for the simulations Simulation 1 and Simulation 2. The
two simulations are explained bellow

  • Simulation 1 : Modification shown in Figure 4.10, current of 50A RMS, fre-
    quency of 50Hz and 38 number of turns

  • Simulation 2: Modification shown in Figure 4.9, current of 50A RMS/phase,
    frequency of 50Hz and 38 number of turns

Table 4.3: Fluxes of phase a for Simulation 1 and Simulation 2. In the case of
Simulation 1 there is no contribution of phase b or phase c. Therefore the calculated
flux is the same with the flux taken from Ansys. The calculated flux comes from
(4.9)

               Simulation                        Fluxes
                                  ΨA (Wb)     from   ΨA calculated
                                  Ansys              (Wb)
               Simulation 1       1.101104           1.101104
               Simulation 2 -     1.4669             1.4681

                                                                                  23
Inductance investigation                                                             24

The next step is to perform the flux calculation in the model shown in Figure 4.11.
First, the influence of 1 coil only (see Figure 4.12) is examined. The total flux caused
from 1 coil only is found to be 1.090834 Wb according to Ansys. Table 4.4 shows
all the fluxes in all slots of phase A from all coils from all phases.

Figure 4.11: coils in the simulated IM 15kW

Figure 4.12: Examining only one coil of phase A

24
Inductance investigation                                                                                    25

Table 4.4: All fluxes that will build via superposition the total flux of phase A
                                            Fluxes from A1+ to A2-

flux type                                             tooth ( slotx slotxx)
             A1+ A2+   A2+ C1-       C1- C2-            C2- B1+               B1+ B2+   B2+ A1-   A1- A2-
flux A1      ΨAA       ΨAA           ΨAA                ΨAA                   ΨAA       ΨAA
flux A2                ΨAA           ΨAA                ΨAA                   ΨAA       ΨAA       ΨAA
MA1A2                  ΨAA           ΨAA                ΨAA                   ΨAA       ΨAA       -ΨAA
MA2A1        -ΨAA      ΨAA           ΨAA                ΨAA                   ΨAA       ΨAA
MC1A1        ΨCC       ΨCC           -ΨCC               -ΨCC                  -ΨCC      -ΨCC
MC1A2                  ΨCC           -ΨCC               -ΨCC                  -ΨCC      -ΨCC      -ΨCC
MC2A1        ΨCC       ΨCC           ΨCC                -ΨCC                  - ΨCC     -ΨCC
MC2A2                  ΨCC           ΨCC                -ΨCC                  -ΨCC      -ΨCC      -ΨCC
MB1A1        - ΨBB     -ΨBB          -ΨBB               -ΨBB                  ΨBB       ΨBB
MB1A2                  -ΨBB          - ΨBB              -ΨBB                  ΨBB       ΨBB       ΨBB
MB2A1        - ΨBB     - ΨBB         - ΨBB              -ΨBB                  -ΨBB      ΨBB
MB2A2                  -ΨBB          -ΨBB               -ΨBB                  -ΨBB      ΨBB       ΨBB

Applying superposition it is obtained

                                 ΨA = 20ΨAA − 8ΨBB − 8ΨCC                                                (4.10)

where
f luxA1 flux caused only from coil1 from phaseA
f luxA2 flux caused only from coil2 from phaseA
M A1A2 mutual flux caused by coil A1 to coil A2
M A2A1 mutual flux caused by coil A2 to coil A1
M C1A1 mutual flux caused by coil C1 to coil A1
M C1A2 mutual flux caused by coil C1 to coil A2
M C2A1 mutual flux caused by coil C2 to coil A1
M C2A2 mutual flux caused by coil C2 to coil A2
M B1A1 mutual flux caused by coil B1 to coil A1
M B1A2 mutual flux caused by coil B1 to coil A2
M B2A1 mutual flux caused by coil B2 to coil A1
M B2A2 mutual flux caused by coil B2 to coil A2

4.1.2       Finding Leakage Inductance
In the previous section, flux distribution was explained for different modified models.
Leakage flux had been eliminated. In this section, leakage flux will be examined.

The initial slot opening is changed as shown in Figure 4.13. A part of the slot
opening is made of air now which will allow a flux to flow. This flux will be the
leakage flux. Rotor bars, like previously, are made of the same material as the core
(µr of 100000). Figure 4.14 shows how the leakage flux will flow through the air in
the slot opening.

                                                                                                            25
Inductance investigation                                                         26

Inductances can be calculated by hand if geometry and material are known. First,
a magnetic circuit is made for the case of one coil as shown in Figure 4.15

Figure 4.13: Modyfied slot opening, light green is the unreal material with µr of
0.00001, one part of the slot opening is made of air with µr of 1

Figure 4.14: B vector through the air in the slot opening

Figure 4.15: Magnetic circuit, relactances of the core material are neglected as the
core material has µr of 100000.

26
Inductance investigation                                                         27

Figure 4.16: Position of the air gap which represents the reluctance RagV

Hand calculations

The reluctances of the circuit in Figure 4.15 are hand-calculated as
                                               Lag
                               Rag =                                          (4.11)
                                        Lstack Arcmean µ0
                                                wslot
                                 RagV =                                       (4.12)
                                          Lstack LagV µ0
                                                   wslot
                           Rslotopening =                                     (4.13)
                                          Lstack Lairopening µ0
where
Rag is the reluctance which represents the path of the flux as a normal component
from the stator to the rotor through the air gap
Rslotopening is the reluctance which represents the path of the leakage flux through
the slot opening
RagV is the reluctance of the air-gap which is located between the end of the unreal
material of the slot opening and the rotor. The part of this air gap is indicated in
Figure 4.16
Lstack is the stack length of the machine
Lag is the length of the air-gap
Arcmean is the mean length of the periphery of the cross section of Rag
wslot is the width of the slot opening
LagV is the length of the path of RagV
Lairopening is the length of the cross section of Rslotopening
µ0 4π · 10−7

If we replace
Lstack = 0.23m,
Lag = 0.45mm,

                                                                                 27
Inductance investigation                                                           28

Arcmean = 99.4mm,
wslot = 0.1mm,
LagV = 0.23mm,
Lairopening = 1mm,
then it is calculated that
Rag = 1.5671 · 104 H −1 ,
Rslotopening = 3.4599 · 105 H −1 ,
and RagV = 1.5043 · 106 H −1

It should be mentioned that RagV is very difficult to represent and en error is ex-
pected. However, an error in the value of RagV will not affect the calculation of slot
leakage inductance.

First, a modified version with Q=18, q=1 and 1 excited coil as shown in Figure 4.9
will be examined. Leakage flux and mutual flux are defined from hand-calculations
as
                                       2
                                      Nturns   N2
                        Lmutual = (          + turns )/NP branches             (4.14)
                                      2Rag    0.5RagV
                                              2
                                            Nturns
                          LLeakage = (                   )/NP branches         (4.15)
                                         0.5Rslotopening
where
Lmutual in this case is the inductance that represents the paths of RagV and Rag
LLeakage is the inductance that represents the flux crossing the slot opening
NP branches is the number of parallel branches per phase
Q is the total number of slots
q is the number of slots per pole per phase

For Nturns = 38 and NP branches = 3, the mutual and leakage inductance become
Lmutual = 15.9975mH and LLeakage = 2.7824mH

Table 4.5 shows the value of the Ansys output L(phaseA, phaseA) (built in Ansys
inductance calculator), the leakage and mutual inductances coming from the field
analysis (using the Ansys field calculator), the self inductance coming from the
division of the total flux (from Ansys result output) with the current (from Ansys
result output) and the hand-calculated inductances for simulation of 1 coil excitation
of 50A DC and no rotor speed.

28
Inductance investigation                                                                           29

Table 4.5: Inductance values. L(phaseA, phaseA) comes from Ansys result output,
Lmutual f ield and LLeakage f ield come from the field analysis, Ls ( Ψi ) comes from An-
sys result output and LLeakage calculated and Lmutual calculated are hand-calculated
                                                      Inductances
                  L(phaseA, phaseA)   Lmutual     Lmutual      LLeakage   LLeakage     Ls ( Ψi )
                                      f ield      calculated   f ield     calculated
Simulation   of   18.66mH             15.884mH    15.99mH      2.768mH    2.782mH      18.66mH
50A DC, Q=18,
q=1, 1 phase/-
coil excited

The case of Q=18, q=1 and 3 excited coils will be now examined. First, (2.37) and
(4.9) are recalled, and the result becomes
                                        1                          1
ΨT mainA = 3ΨAA − 1ΨBB − 1ΨCC = ΨmainA − (ΨmainB + ΨmainC ) = (Lh + Lh )isa
                                        3                          3
                                                                     (4.16)
and it is also known that
                                                      4
                         Ψa = ΨT mainA + ΨleakageA = ( Lh + Lleakage )isa                   (4.17)
                                                      3

Combining (4.16) and (4.17), the total flux of phase a can be expressed as

         Ψa = Lh isa + 1/3Lh isa + Lleakage isa = (Lh + Lleakage )isa + M isa               (4.18)
and the mutual inductance between 2 phases can be expressed as
                                                    1
                                                 M = Lh                                     (4.19)
                                                    3
where
ΨT mainA is the total main flux of phase a
ΨleakageA is the leakage flux of phase a
ΨmainA is the main flux of a coil of phase a
ΨmainB is the main flux of of a coil of phase b
ΨmainC is the main flux of of a coil of phase c
M is the mutual inductance between 2 phases
Lh is the main inductance of phase a caused only from the influence of phase a
Lleakage is the leakage inductance of phase a, b or c

It is reminded that in the examples where flux distribution was examined, leakage
flux was eliminated which means that the flux of one phase was the main flux. Now
the leakage inductance caused by leakage flux is added to the equation.

Table 4.6 shows the Ansys output L(phaseA, phaseA), L(phaseA, phaseB) (built in

                                                                                                   29
Inductance investigation                                                                     30

Ansys inductance calculator), the hand-calculated leakage inductance and the self
inductance coming from the division of stator flux with the stator current. Maverage
represents the mean value of the Ansys outputs L(phaseA, phaseB), L(phaseA, phaseC)
and L(phaseC, phaseB).

Table 4.6: Inductances for case of Q=18, 1 excited phase with current of 50ADC.
                                                      Inductances
                                                                                      Ψ
                    L(phaseA, phaseA) L(phaseA, phaseB) Maverage    LLeakage hand− Ls i A
                                                                                       sa
                                                                    calculated
Simulation     of   18.66mH          5.2546mH          -5.288mH     2.768mH        23.94mH
50A, f=50Hz, 3
excited coils, 0
rotor speed

If it is assumed that the outputs from Ansys inductance matrix can be interpreted as

                               L(phaseA, phaseA) = Lh + Lleakage                          (4.20)
and

                                    L(phaseA, phaseB) = −M                                (4.21)
then the self inductance can be expressed as

                        Ls = L(phaseA, phaseA) − L(phaseA, phaseB)                        (4.22)

Using (4.22) for Ls calculation an error of 0.0254mH is found, which corresponds to
0.1% error in Ls

It is also observed that L(phaseA, phaseB), L(phaseA, phaseC) and L(phaseC, phaseB)
have a very slightly different value, the reason of this deviation is unknown. Replac-
ing M with Maverage , (4.22) gives an error of 0.008mH, which corresponds to 0.033%
error in Ls. If the error is accepted then it can be concluded that

                                    L(phaseA, phaseB) = −M                                (4.23)
and
                               L(phaseA, phaseA) = Lh + Lleakage                          (4.24)

According to (4.19) and (4.24), the leakage inductance Lleakage is then calculated as

                    Lleakage = L(phaseA, phaseA) + 3L(phaseA, phaseB)                     (4.25)

30
Inductance investigation                                                                     31

The leakage inductance that is calculated according to (4.25) is 2.796mH which gives
an error of 1.01% in the estimation of leakage inductance.

With the same strategy as before, the case of Q=36, q=2 with the same materials as
before is examined. From (4.23) and (4.10) the mutual inductance can be expressed
as

                                                              8
                           M = −L(phaseA, phaseB) =              Lh                      (4.26)
                                                              20

and combining (4.23), (4.23) and (4.26) the leakage inductance can be expressed as

                                                      20
               Lleakage = L(phaseA, phaseA) +            L(phaseA, phaseB)               (4.27)
                                                       8

Table 4.7 shows the calculated leakage inductance from field analysis (using Ansys
field calculator), the total leakage inductance from hand-calculation (the number of
stator slots are different now) and the two Ansys output inductances (built in Ansys
inductance calculator) for the simulation of 50A RMS current and 50Hz.

Table 4.7: Inductances from Ansys result output, hand-calculated leakage induc-
tance and leakage inductance from the field analysis (using Ansys field calculator)
                                                   Inductances
                    L(phaseA, phaseA) L(phaseA, phaseB) Lleakage calculated Lleakage from field
                                                                            analysis
 Simulation    of   58.124mH          -21.207mH         5.1065mH            5.52mH
 50A, f =50Hz,
 3 excited coils,
 Q=36 and q=2

4.1.3     Adding saturation

Figure 4.17 shows how the inductances will oscillate in case of a non-linear core
material. The model of Q=18, q=1 will be used for the simulations. The stator slot
modification is the same as in the previous chapter.

                                                                                             31
Inductance investigation                                                           32

Figure 4.17: Indactances from Asnys output data when non-linear M700 core
material is used. Model of Q=18, q=1 and 3 excited coils where used with excitation
of 50A RMS and 50Hz

According to [10] the component of the 2nd harmonic of the mutual inductance is
also contributing to the fundamental flux. Considering the second harmonic of the
inductances the equation for the self inductance calculation can be expressed as

                       Ls = LAADC + MDC − LAA2 − M2                            (4.28)

If the 2nd harmonic is not taken into account then the self inductance is expressed as

                               Ls = LAADC + MDC                                (4.29)
and the leakage inductance is expressed as

                              Lsl = LAADC − 3MDC                               (4.30)
where
LAA is the sum of the main inductance of phase a and the stator leakage inductance
of phase a
M is the the mutual inductance between 2 phases
LAADC is the mean value of LAA
MDC is the mean value of M
LAA2 is the second harmonic magnitude of LAA
M2 is the second harmonic magnitude of M
Lsl is the leakage stator inductance of one phase

It should be noted that the calculation of leakage inductance has been made only
according to (4.30)

Simulations with a current sweep from 5A to 45A RMS at 50Hz and no rotor speed

32
Inductance investigation                                                                                         33

have been made. More specifically, the following simulations have been made:

        • Simulation 1: linear core material, no rotor bars

        • Simulation 2: steel M700 core material, no rotor bars

        • Simulation 3: steel M700 core material, aluminium rotor bars without eddy
          effect in them

        • Simulation 4: steel M700 core material, no rotor bars and the special flux
          blocker material covers all the volume of the stator slot opening so leakage
          inductance is eliminated

Figure 4.18 shows the leakage inductance (for Simulation 1, Simulation 2 and Sim-
ulation 4) by means of calculation according to (4.30) and the measured leakage
inductance from the field analysis. It is observed that (4.30) is not giving very good
estimation of leakage inductance in the cases of M700 as core material.

       3.5
                                                             L   from inductance DC evaluation, linear core
                                                             L   field, linear core
         3
                                                             L   from inductance DC evaluation, M700 core
                                                             L   field, M700 core
       2.5                                                   L   from inductance DC evaluation, M700 core, case of full applied flux blocker
                                                             L   field, M700 core, case of full applied flux blocker (Lsl=0)

         2
[mH]

       1.5
L

         1

       0.5

         0

       -0.5
              5   10   15   20    25     30   35   40   45
                            RMS current [A]

Figure 4.18: Comparison of calculated leakage inductance according to (4.30) and
the leakage inductance measured from the field analysis (from Ansys field calculator)

Figure 4.19 shows the calculated self inductance using (4.28), (4.29) and the one
that comes from Ψa/ia, where Ψa and ia are taken from the Ansys result output.

                                                                                                                 33
Inductance investigation                                                                               34

          24

          22
Ls [mH]

          20
                   Ls Psi/i linear core with Lsl
          18        Ls from inductance DC eval linear core with Lsl
                   Ls Psi/i M700 core with Lsl
          16
                    Ls from inductance DC eval linear core with Lsl
          14
               5         10                15               20              25          30   35   40        45
                                                                      RMS current [A]
          22

          20
Ls [mH]

          18
                   inductance DC and 2nd harm evaluation
          16       inductance from Psi/i
                   inductance DC evaluation
          14
               5         10                15               20              25          30   35   40        45
                                                                      RMS current [A]
          20

          18
Ls [mH]

          16
                   inductance DC and 2nd harm evaluation
          14       inductance from Psi/i
                   inductance DC evaluation
          12
               5         10                15               20              25          30   35   40        45
                                                                      RMS current [A]

Figure 4.19: Upper plot: self inductance calculated from (4.29) and Ψa/ia for
the cases of Simulation 1 and Simulation 2. Middle plot: self inductance calculated
from (4.28) (blue dash line), (4.29) (red dash-dot line with cross) and Ψa/ia (green
solid line) for the cases of Simulation 2. Lower plot: self inductance calculated from
(4.28) (blue dash line), (4.29) (red dash-dot line with cross) and Ψa/ia (green solid
line) for the cases of Simulation 2.

34
Inductance investigation                                                                        35

Testing equations in IM at no load

It is desired to see what results the equations (4.28) and (4.29) will give in case of
calculating self inductance at no load operation. Now, the model is the initial IM
machine with Q=36, q=2, core material of M700 and rotor bars made of aluminium.
No load simulations at 50Hz, 1000rpm and a current sweep from 5A to 40A RMS are
made. In Figure 4.20 is shown the self inductance estimated from Ψa/ia (coming
from the Ansys result output) and the self inductance calculated from (4.28) and
(4.29) (making use of the inductances built in Ansys inductance calculator).

                55

                                            Ls from a /ia from Ansys
                                            Ls inductance DC and 2nd harmonic evaluation
                50                          Ls inductance from only DC evaluation

                45

                40
      Ls [mH]

                35

                30

                25

                20
                     5   10   15       20          25           30          35             40

                                     RMS current [A]

Figure 4.20: Solid green line: self inductance calculated from Ansys outputs ψa,
ia. Blue dash line: self inductance calculated from 4.28. Red dash-dot line: self
inductance calculated from 4.29

Figure 4.21 shows the flux of the phase a from Ansys and the one obtained using
(2.37) for the no load simulation at 40A RMS. It is observed that (2.37) does not
reflect to the flux with a great precision. The reason for this is unknown.

                                                                                                35
Inductance investigation                                                                         36

                    1.5

                                                                          a
                                                                              from Ansys

                                                                          a
                                                                              calculated

                      1

                    0.5
        flux [Wb]

                      0

                    -0.5

                     -1

                    -1.5
                      1.465   1.47   1.475   1.48          1.485   1.49   1.495            1.5

                                                    time [s]

Figure 4.21: Black solid line: flux of phase a from Ansys result output. Blue
dash-dot line: calculated flux using 2.37 (making use of the inductances built in
Ansys inductance calculator)

36
Power balance check                                                                  37

4.2      Power balance check
A power unbalance have been observed after processing Ansys output data. This
causes an error to the flux, torque and voltage equations we may use for parameter
identification.

Induced voltage is an output from Ansys simulation which is calculated from Ansys
by differentiate the flux,
                                                dψa
                                         ea =                                    (4.31)
                                                Tstep
where
ea is the induced voltage of phase a
ψa is the flux of phase a
Tstep is the time step

That means that ea will be delayed by half of a time step. A phase correction of ea
will lead re-express ea as

                            eanew =| eaold | ej(θold +θcorrection )              (4.32)
where:
eaold is the induced voltage before the correction
eanew is the induced voltage after the correction
θold is the angle of induced voltage before the correction
and
                                                     Tstep
                                 θcorrection = 0.5         2π                    (4.33)
                                                      T

where T is the period

Power balance at locked rotor test

It is desired to see how the correction angle from (4.33) will correct the power balance
in the locked rotor test. A locked rotor test is carried out at frequency of 10Hz and
a current sweep from 10 to 30A RMS. Table 4.8 shows the power missing in case of
not using the correction angle. Table 4.9 shows the power balance in case of using
the correction angle.

Power balance at load test

                                                                                     37
Power balance check                                                                            38

Table 4.8: Discrepancy in power balance in case of not correcting the induce voltage
                         Power balance at locked rotor without corrected angle
Current sweep   10A RMS           15A RMS           20A RMS            25ARMS        30A RMS
Discrepancy     4.2W              9.6W              17.1W              26.8W         38.5W

Table 4.9: Discrepancy in power balance in case of correcting the induced voltage.
                          Power balance at locked rotor with corrected angle
Current sweep   10ARMS            15ARMS            20ARMS             25ARMS        30ARMS
Discrepancy     0.30W             0.69W             1.25W              1.95W         2.80W

Power balance is not satisfied in case of a load test even if (4.33) is used. The exact
reason for this is unknown.

In case of a load test torque unbalance is observed between the Torque from Ansys
output and the calculated electromagnetic torque (see(2.5)). This torque unbalance
could be a possible reason for the power unbalance that is observed for the case of
a load test.

Three simulation with voltage excitation at 220V RMS, 50Hz and slip frequency
of 0.5Hz, 1Hz and 1.5Hz are made. Table 4.10 shows the torque from Ansys and
the calculated electromagnetic torque for the three operation points. Table 4.11
shows the error in the power balance for the three operation points. It is noted
that in the equation of power balance the torque from Ansys is used and not the
electromagnetic calculated torque.

Table 4.10: Comparison between torque from Ansys and the calculated electro-
magnetic torque (Te stands for the calculated electromagnetic torque)

      Torque comparison in Voltage excitation of 220V RMS and 50Hz
 slip frequency           0.5Hz                      1Hz                         1.5Hz
 Te calculated            89.2787Nm                  166.7245Nm                  240.2055Nm
 Torque Ansys             88.3027Nm                  164.7277Nm                  236.9517Nm
 Discrepancy              0.976Nm                    1.9968Nm                    3.2538Nm

38
Power balance check                                                               39

Table 4.11: Power balance for the three operation points. Power in is the input
electrical power, Power out stands for mechanical output power, Power loss stands
for the total losses, Discrepancy stands for the unbalance between the Power in and
the summation of the mechanical output power and total losses. The calculation of
the output mechanical power uses the torque from Ansys and not the electromagnetic
calculated torque

         Power balance in Voltage excitation of 220V RMS and 50Hz
 slip frequency       0.5Hz                  1Hz                 1.5Hz
 Power in             9492.7W                17785.3W            25770W
 Power out            9150.4W                16905.2W            24080W
 Power loss           358.5W                 890.7W              1722.7W
 Discrepancy          16.14W                 10.65W              32.97W

Torque calculation from field analysis

Some efforts have been made in order to calculated the torque from field analysis.
First, the Maxwell stress tensor needs to be calculated,

                                          2
                                  Bt Bn Bmag
                               s=      −     δtn                               (4.34)
                                   µ0    2µ0

where
s is the Maxwell stress tensor
Bt is the tangential component of B vector
Bn is the normal component of B vector
µ0 4π10−7
Bmag is the magnitude of B vector
δ is the Kronecker’s delta.

In the case where only the tangential force is needed the second term is ignored.
The Maxwell stress tensor equation is simplified

                                          Bt Bn
                                     s=                                        (4.35)
                                           µ0

The line which was used in order to sample the points of interest for the calculation
of the Maxwell stress tensor was taken according to Figure 4.22.

                                                                                  39
Power balance check                                                                40

Figure 4.22: Used line (purple) to sample the B vector in order to calculate the
Maxwell stress tensor. Some parts of the machine like a part of the rotor and the
stator are hidden in order to make the line more visible

The torque can be expressed with the Maxwell stress tensor and the geometry of
the machine as

                            T orque = saverage Aouter rrotor                    (4.36)

where:
saverage is the average stress across the line
Aouter is the outer surface of the rotor (outer periphery times stack length)
rrotor is the radius of rotor

The simulation of voltage excitation at 220V RMS, frequency of 50Hz and speed
of 980rpm will be examined. Figure 4.23 shows the torque that Ansys provides
as output and the calculated torque according to (4.36) and sampling according
to Figure 4.22. Moreover, Figure 4.23 shows the calculated torque using an extra
smooth operator in Ansys field calculation. The position of this operator is shown
in Figure 4.24.

Table 4.12 shows the average torque coming from Ansys result output, the average
calculated torque according to (4.36) and the average calculated torque using an
extra smooth operator in Ansys field calculation.

40
Power balance check                                                                                     41

                    180

                             Torque from Ansys
                             Torque calculated SMOOTH
                    175
                             Torque calculated

                    170
      Torque [Nm]

                    165

                    160

                    155

                    150

                    145
                     1.955    1.96     1.965    1.97    1.975         1.98   1.985   1.99   1.995   2
                                                            time[s]

Figure 4.23: Green solid line: Torque from Ansys. Blue dashed-dot line: calculated
torque. Red dashed line: calculated torque using the smooth operator

                                                                                                        41
Power balance check                                                                 42

Figure 4.24: Position of smooth operator in Ansys field calculator

Table 4.12: Average value of the torque for 220V RMS, 50Hz and 980rpm.
T orqueAnsys stands for the average torque coming from Ansys result output.
T orquecalculated stands for the average torque that is calculated according to (4.36).
T orquesmooth stands for the average torque that is calculated according to (4.36) but
using the smooth operator from Ansys field calculator

                       Torque kind           value
                       T orqueAnsys          164.7111Nm
                       T orquecalculated     159.8686Nm
                       T orquesmooth         165.5841Nm

42
Rotor resistance                                                                  43

4.3      Rotor resistance
In this chapter an estimation of rotor resistance is made. The rotor resistance is
estimated from a locked rotor test and from field analysis.

Moreover, the rotor resistance can be calculated considering the material and geom-
etry dimensions. Equations (2.47), (2.48), (2.54) are recalled, giving

                    ρal = 28.2 · 10−9 · (1 + 4.3 · 10−3 · (Tal − 20))          (4.37)
                                  Rbar = ρal La /Abar                          (4.38)
where
Tal is the temperature of the aluminium in Celcius
La is the length of the bar
Abar is the cross section of the bar in the case of current in z direction

Considering the temperature inside the aluminium to be 30, the resistivity of alu-
minium is found to be 2.9413· 10−8 m. However, in this project a resistivity of
3.0252 · 10−8 m is used. In addition, the cross section of the rotor bar Abar is mea-
sured from Ansys and is found to be 76.41786574 · 10−6 m2 . The stack length LA is
0.23m. Therefore (4.38) gives a value of 9.1052 · 10−5 which is the resistance of one
rotor bar.

The coefficient to connect the rotor equivalent resistance to stator equivalent is
calculated according to (2.51) which is recalled
                                           3n2s qs2 k12 poles2
                              coefreqS =                                       (4.39)
                                                  Qr c2s
where
Qr is the number of total bars
poles is the number of total poles
ns is the number of turns per coil
qs is the number of slots per phase per pole pair
k1 is the harmonic winding factor
cs is the winding factor which depends on the way the phases are connected and the
parallel branches

If Qr = 38, poles = 6, ns = 38, qs = 1, cs = 3 and k1 = 0.9659, then coefreqS =
1.6582 · 103 . The stator equivalent rotor resistance is calculated as

                       Rr = Rbar coefreqS = 0.1509799            (Ω)           (4.40)

It should be noted that the end ring is not modelled in the simulations.

                                                                                  43
Rotor resistance                                                                  44

4.3.1    Rotor resistance from locked rotor test

Some efforts have been made in order to calculate the rotor resistance from the
locked rotor test. Simulations have been carried out at 10Hz, 0rpm and a current
sweep from 10 to 30 A RMS. It should be noted that in case of current excitation
Ansys does not consider the stator resistance

Two approaches have been made. One using τ model and one approach using k
model.

τ model approach

Figure 4.25 shows the simplified equivalent circuit used for the calculation of rotor
resistance. The value of mutual inductance is unknown and difficult to estimate.
Therefore an aproximation between the stator current and the rotor current is made

                                     iˆr = 0.95iˆs                             (4.41)

The rotor resistance is calculated from

                                           Psolid
                                    Rr =                                       (4.42)
                                            3 ˆ2
                                            2
                                             ir
where
Rr rotor resistance in τ model
iˆr is the pick of the fundamental of the rotor current
iˆs is the pick of the fundamental of the stator rotor current
Psolid is the average solid losses taken from Ansys result output

Figure 4.26 shows the results of Rr according to (4.42)

Figure 4.25: simplified τ circuit for the case of current excitation

44
Rotor resistance                                                                                     45

                   0.1626

                   0.1625

                   0.1624

                   0.1623
        Rr [ohm]

                   0.1622

                   0.1621

                    0.162

                   0.1619

                   0.1618

                                                                                        Rr
                   0.1617
                         10   12   14       16    18       20       22   24   26   28        30

                                                       is RMS [A]

Figure 4.26: Rr calculated according to (4.42) for a current sweep of 10-30A RMS

k model approach

Figure 4.27 shows the k model. Now the resistance of the rotor in k model will
be estimated. It should be noted that in case of current excitation Ansys does not
consider the stator resistance

The total impedance of the circuit of Figure 4.27 is

                                        Zcircuit = ZLsigna + ZLM //RR                             (4.43)

Zcircuit is the total impedance from the circuit
ZLsigna is the impedance of the leakage inductance in k model
ZLM is the impedance of magnetization inductance
RR is the rotor resistance in k model

The total impedance can be also expressed from the relation between stator voltage
and stator current
                                                           es
                                                  Zei =                                           (4.44)
                                                           is
where
Zei is the total impedance from the ratio of voltage over current

                                                                                                     45
Rotor resistance                                                             46

es is the induced voltage
is is the stator current

Setting (4.43) and (4.44) equal, it is obtained that
Rotor resistance                                                                                47

                    0.1495

                                                              RR for current of 10 A RMS
                                                              RR for current of 15 A RMS
                                                              RR for current of 20 A RMS
                                                              RR for current of 25 A RMS
                     0.149                                    RR for current of 30 A RMS

                    0.1485
        R R [ohm]

                     0.148

                    0.1475

                     0.147
                          25   30   35   40      45     50        55         60            65
                                              LM [mH]

Figure 4.28: RR values at locked rotor test during a sweep of LM .

                     3.36

                     3.34

                     3.32

                      3.3

                     3.28
          L [mH]

                     3.26

                     3.24

                     3.22
                                                        Lsigma for current of 10 A RMS
                                                        Lsigma for current of 15 A RMS
                      3.2                               Lsigma for current of 20 A RMS
                                                        Lsigma for current of 25 A RMS
                                                        Lsigma for current of 30 A RMS
                     3.18
                         25    30   35   40     45      50        55         60          65
                                              LM [mH]

Figure 4.29: Lsigma values at locked rotor test during a sweep of LM .

                                                                                                47
Rotor resistance                                                                 48

4.3.2    Rotor resistance from field analysis

Some efforts have been made in order to estimate the rotor resistance from field
analysis at load simulation.

First, let us begin with the torque equation in the DQ frame
                                            3np
                                     Te =       ΨR iq                        (4.48)
                                             2

It is known from the theory that

                                        ΨR = bψr                             (4.49)
and
                                                ir
                                         iR =                                (4.50)
                                                b

Equations (4.50) and (4.49) give

                                      ΨR iR = ψr ir                          (4.51)

In addition, it is known from the theory that the rotor current can be expressed as

                                     ir Rr = ψrwslip                         (4.52)
                                     ir = ψrwslip /Rr                        (4.53)

Combining (4.48), (4.51) and (4.53), Rr can be expressed in terms of Te and ir

                                                 Te
                                      Rr =   np ·3/2·i2r
                                                                             (4.54)
                                               wslip

where
Te is the electromagnetic torque
np is the number of pole pairs
ir is the rotor current in τ model

48
Rotor resistance                                                                          49

wslip is the slip angular speed
Rr is the Rotor resistance in τ model
ψr is the rotor flux in τ model
ΨR is the rotor flux in k model
iR is the rotor current in k model
b is the important ratio constant

The rotor current was taken from the field analysis. Figure 4.30 shows the current in
the z direction of a rotor bar for the simulation of voltage excitation at 220V RMS,
50Hz and slip frequency of 1Hz. This current is the rotor bar current in the rotor
side. In order to obtain the equivalent rotor current at the stator side, coefficient
according to (2.51) is used

                                          ir = irr coefieqS                            (4.55)
where
irr is the rotor current in the rotor side
ir is the funtametal of the rotor current in the stator side
coefieqS is the coefficient to connect the rotor equivalent current to stator equivalent

                       600

                                                                          Iz bar

                       400

                       200
         current [A]

                         0

                       -200

                       -400

                       -600
                          0.8   1   1.2             1.4       1.6   1.8            2
                                                  time [s]

Figure 4.30: Rotor current in the rotor side for the simulation of 220V RMS, 50Hz
and 1Hz of slip frequency. The time of sampling is 1 second which is the slip period
of the rotor current

Four simulation have been made with voltage excitation of 220V RMS, 50Hz and
a slip frequency of 1, 1.5, 2, and 2.5 Hz. Table 4.13 shows the rotor resistance

                                                                                          49
Rotor resistance                                                                           50

according to (4.54) using rotor current from field sampling.

Table 4.13: Te was calculated according to (2.5), ir is the stator equivalent rotor
current (the fundamental component) from the field analysis and Rr is calculated
according to (4.54)
                Voltage excitation 220V RMS, 50Hz and   3 slip frequency points
 slip frequency      1Hz               1.5Hz             2Hz                 2.5Hz
 ir                  39.1998 A         57.954 A          76.0441 A           93.4207 A
 Rr                  0.15156 ohm       0.15167 ohm       0.15156 ohm         0.15168 ohm
 Te                  166.7973 Nm       243.2201 Nm       313.8533 Nm         379.2279 Nm

50
Stator leakage inductance                                                           51

4.4     Stator leakage inductance
In this chapter stator leakage inductance is approximated. The material of the core
is changed to a steel with constant relative permeability of 100000. In this way the
effect of saturation will not affect the simulation. Two approaches have been made

Approach 1: No load test

In this approach, the leakage inductance is estimated from a no load test using the
Ansys inductance matrix. If a linear core material is used then Ansys inductance
matrix can be useful as shown in chapter 4.1. A no load test is been carried out at
frequency of 50Hz, rotor speed of 1000rpm and phase current of 32 A RMS. Table
4.14 shows the Ansys outputs L(P haseA, P haseA), L(P haseA, P haseB) and the
calculated inductance from (4.27)

Table 4.14: Calculation of leakage inductance Lσλ calculated from no load test
and using the data from Ansys inductance matrix and the equations from flux
distribution (see(4.27 ))

 Simulation           L(P haseA, P haseA)   L(P haseA, P haseB)    Lσλ calculated
 No load simulation   39.458mH              15.5363mH              0.617mH
 at 32A RMS, 50Hz
 and 1000rpm rotor
 speed

Approach 2: Making the rotor a flux blocker

In this approach the rotor has been replaced by an unreal material with relative
permeability of 0.00001. In this way all the flux is flowing from the paths of the
leakage inductance. A simulation has been made at a current of 32A RMS and
frequency of 50Hz. Table 4.15 shows the the Ansys outputs L(P haseA, P haseA),
L(P haseA, P haseB), the calculated inductance from (4.27) and the calculated self
inductance from the division of ψa /ia (ψa and ia are taken from Ansys result out-
put).

The leakage inductances that were calculated by approach 1 and approach 2 are
not identical. However, the value is small and in case of approach 1 an error would
affect the value of the estimated leakage inductance a lot.

                                                                                    51
Parameter estimation                                                                        52

Table 4.15: Calculation of leakage inductance in the case of replacing the rotor
with the material of very low permeability. All the flux is in principle the measured
leakage flux which gives the leakage inductance
 Simulation      L(P haseA, P haseA) L(P haseA, P haseB) Lσλ calculated   Ls (ψa /ia )
 Simulation      0.8482mH            16nH                0.8482mH         0.8482mH
 at 32A RMS,
 50Hz

4.5        Parameter estimation
In this chapter a parameter estimation will be described.

4.5.1       Processing the data
The inputs used for the parameter estimation in this chapter are

     • ψs : the fundamental component of the stator flux (taken from Ansys result
       output) in xy coordinates

     • is : the fundamental component of the stator current (taken from Ansys result
       output) in xy coordinates

     • ωslip : the slip angular speed

     • iˆr : the magnitude of the stator equivalent rotor current from the field analysis
       (see (4.55) )

     • b : the important ratio constant (see (2.6)). A guess of b is made in order to
       estimate the possible parameters.

If the rotor leakage inductance Lrλ is expressed as a ratio a of the mutual inductance
Lm , then the important ratio constant b can be expressed as

                                        Lrλ = aLm                                        (4.56)
                                 Lm          Lm       1
                          b=            =          =                                     (4.57)
                               Lm + Lrl   Lm + aLm   1+a

and in this case a guess of b means a guess of a.

First, the electromagnetic torque Te is estimated from
                                               np
                                      Te = 3      =ψs∗ is                                (4.58)
                                               2

52
You can also read